Lean AI Glossary

Eighteen defined terms from the Lean AI methodology. Use these to anchor a conversation with the team, a vendor, or yourself before spending money on AI tools. Each term links to the related tool, service, or pillar page where you can go deeper.

New to the method? Start with The Lean AI Method or the AI Readiness pillar.

Index

90-Day Implementation Sprint
AI Advisory Retainer
AI Champion
AI Governance
AI Necessity Test
AI Operating Model
AI Pilot
AI Readiness
AI Readiness Audit
AI ROI
AI Skill Gap
AI Strategy Workshop
AI Use Case
Digital Gemba (Walk)
Lean AI Method
Process Friction
Three-Question Filter
Value Stream Mapping (for AI)

90-Day Implementation Sprint

A fixed-scope, fixed-price AI build that ships a single proven use case in 90 days. Includes process mapping, build, integration, training, and a measured ROI handover. The Sprint replaces the open-ended 12-month consulting contract: there is one outcome, one deadline, and one number on the invoice. At LeverageAI the Sprint is £30,000 and is the third rung in the offer ladder.

AI Advisory Retainer

Ongoing monthly engagement after a Sprint completes. The team has a working AI use case; the Retainer keeps it healthy, tunes performance, expands scope, and helps prioritise the next build. At LeverageAI the Retainer is £3,000/month and is the fourth rung in the offer ladder.

AI Champion

The internal team member who owns AI adoption inside a business. Not a job title — a role. The Champion holds the use case shortlist, runs the pilot, surfaces blockers, and trains the rest of the team. Without a named Champion, AI projects stall the moment the external partner leaves.

AI Governance

The set of policies, controls, and review processes that determine how AI gets built, deployed, and monitored inside a business. Covers data access, model selection, bias review, audit logging, human-in-the-loop sign-off, and incident response. Most mid-market businesses don't need a 60-page policy document — they need three pages and someone with the authority to enforce them.

AI Necessity Test

A free 8-minute diagnostic that scores whether a specific business process actually needs AI, or whether a simpler change (process, training, a spreadsheet) would deliver more ROI. Built from Lean Six Sigma principles. Returns a readiness score plus the next recommended step.

AI Operating Model

How a business organises people, processes, technology, and decisions around AI. Covers who owns AI use cases (centralised vs federated), how budgets flow, how priorities get set, and how the team measures impact. A clear operating model is the difference between AI as a one-off project and AI as a repeatable capability.

AI Pilot

A short, time-boxed AI proof-of-concept with explicit success criteria, a defined business problem, and a target metric. Typical pilot length is 4-12 weeks. A pilot is not a 'platform purchase' — it's a test of whether AI moves the metric in this specific process. Pilots that don't move the metric are killed, not extended.

AI Readiness

The state of having the prerequisites in place to adopt AI without it failing for non-AI reasons. The four pillars are process readiness (workflows are defined), data readiness (data is accessible and clean), people readiness (skills and culture exist), and strategy readiness (clear business outcomes). AI readiness comes before AI maturity, which is what happens after you've shipped your first pilot.

AI Readiness Audit

A structured 14-day engagement that scores a business on the four readiness pillars, runs digital Gemba walks across the highest-value workflows, prioritises a shortlist of AI use cases, and produces a 90-day implementation roadmap. At LeverageAI the Audit is £7,500 and credits in full toward a subsequent Sprint.

AI ROI

Return on investment specific to AI projects. Calculated as (annualised value created - total cost of ownership) ÷ total cost of ownership. 'Total cost of ownership' includes licences, build cost, integration, training, ongoing maintenance, and the cost of changing the underlying process. Most AI ROI calculations skip the last two and overstate the return.

AI Skill Gap

The distance between the skills a team currently has and the skills needed to operate AI in their workflow day-to-day. Includes prompting, model selection, evaluating outputs, knowing when to override, and basic data hygiene. Closing the gap is usually faster and cheaper than people expect — but ignoring it is the single most common cause of pilot failure.

AI Strategy Workshop

A half-day workshop that aligns the leadership team on which business problems are worth solving with AI, what success looks like, and what the first 90 days should produce. The Workshop is the bridge into the LeverageAI ladder — it's where the use case shortlist gets sketched and the first audit gets scoped. £3,000 at LeverageAI.

AI Use Case

A specific, scoped business problem where AI is a candidate solution. Phrased in a single sentence: 'We want to [verb] [thing] for [team] so that [outcome] measured by [metric].' If the sentence doesn't fit, the use case isn't ready. Most failed AI projects are use cases that were never properly defined.

Digital Gemba (Walk)

A screen-share walkthrough of how work actually happens, step by step, click by click — adapted from the Lean manufacturing Gemba (go to the place where the work happens). For knowledge work the 'place' is the screen. A digital Gemba surfaces the friction, workarounds, and unspoken rules that don't appear in any process diagram. It's the foundation of any honest AI readiness assessment.

Lean AI Method

The methodology LeverageAI uses to help mid-market businesses adopt AI without the 95% failure rate. Five principles: process first, technology second; pilot before platform; skin in the game (90-day sprints, not 12-month contracts); skill the team, don't replace it; measure ROI from day one. Built from a decade of Lean Six Sigma work applied to AI implementation.

Process Friction

Measurable inefficiency in a workflow: rework, waiting, hand-offs, duplicated effort, manual reconciliation, context-switching. Friction is what AI candidates look for. A process with low friction is a bad AI candidate (there's nothing for AI to do); a process with high friction may be an AI candidate, but is often a process-design problem first.

Three-Question Filter

A 3-question filter that decides whether a process is worth automating with AI. The questions: Does this process repeat at meaningful volume? Is the cost of getting it wrong tolerable? Is there a clear measure of success? If any answer is 'no', the process isn't an AI candidate yet — fix the gap first.

Value Stream Mapping (for AI)

The Lean technique of mapping every step from customer trigger to value delivered, with time and cost data on each step. Applied to AI, it surfaces where AI can move a real metric versus where it would only deliver cosmetic value. A value stream map is the cheapest way to kill bad AI ideas before they cost real money.